Involvement of phosphatidylserine, αvβ3, CD14, CD36, and complement C1q in the phagocytosis of primary necrotic lymphocytes by macrophages

Authors


Abstract

Objective

Uningested dead cells may be an important source of autoantigens and may trigger autoimmune diseases such as systemic lupus erythematosus (SLE). Multiple receptors involved in the clearance of apoptotic cells have been described; however, little is known about the receptors and ligands involved in uptake of necrotic cells that release autoantigens as well.

Methods

The uptake of autologous necrotic peripheral blood lymphocytes into human monocyte-derived macrophages was qualitatively and quantitatively monitored by confocal microscopy and 2-color flow cytometry, respectively. Blocking experiments were performed to examine the receptors and molecules involved in the phagocytosis of necrotic cells. Cytokine secretion by lipopolysaccharide-activated monocytes and macrophages was determined by enzyme-linked immunosorbent assay.

Results

Phosphatidylserine, which was exposed on necrotic as well as apoptotic cells, promoted the recognition and removal of primary necrotic lymphocytes. Several macrophage receptor systems, including the thrombospondin–CD36–αvβ3 complex, CD14, and the complement component C1q, contributed to the engulfment of necrotic cells. Necrotic peripheral blood lymphocytes slightly increased the lipopolysaccharide-induced secretion of interleukin-10 and reduced the secretion of tumor necrosis factor α in monocytes and macrophages.

Conclusion

Our results indicate that at least some of the receptors and adaptors mediating the uptake of apoptotic cells are also involved in the clearance of necrotic cells. Hence, necrotic cells engage phagocyte receptors such as CD36, which mediate antiinflammatory signals from apoptotic cells. Necrotic cells consequently also have the potency to provide antiinflammatory signals to phagocytes; however, these signals may be overridden by proinflammatory factors released during necrosis. These findings have implications regarding the etiopathogenesis of autoimmune diseases such as SLE, in which impaired clearance of dead cells may foster autoimmunity by the release of potential autoantigens.

Apoptotic cells, which maintain their membrane integrity for a certain time, have to be cleared quickly and efficiently to prevent the release of tissue-damaging intracellular constituents and subsequent inflammation. Also, necrotic cells need to be removed efficiently to limit organ damage and enable timely tissue repair. Apoptotic cells are normally cleared via an antiinflammatory pathway (1). Many receptors and adaptor molecules have been shown to contribute to the recognition and uptake of apoptotic cells by phagocytes (for review, see refs. 2 and3). Among these are collectin receptors, calreticulin/CD91, Fcγ receptors, c-Mer, the β2-glycoprotein I receptor, integrins such as αvβ3, lectins, CD14, ABC transporters, scavenger receptors including thrombospondin (TSP) receptor CD36, and a putative, not-yet-identified phosphatidylserine (PS) receptor, which appears to be recognized and blocked by the monoclonal antibody (mAb) 217G8E9 (4–10).

Most of these receptors do not directly bind to apoptotic cells; rather, the dying cells are engaged via bridging molecules (for review, see ref. 3). Some of these adaptor proteins have been identified. They often represent regular plasma components. The TSP receptor (CD36) and the vitronectin receptor αvβ3 cooperate in binding TSP-1, which interacts with yet-unknown sites on apoptotic cells (2, 11, 12). Collectins, such as mannose-binding lectin and the collectin-like protein C1q, have been shown to bind to late apoptotic cells and drive ingestion through interaction with calreticulin and CD91 on the phagocyte in vitro. Surfactant protein D (SP-D) has been shown to enhance clearance of apoptotic polymorphonuclear cells from the naive murine lung (13), but neither SP-A nor C1q did so, whereas in other systems C1q is important in the clearance process as an opsonizing agent (4, 14–17).

The surface exposure of PS represents an important and evolutionarily conserved recognition signal for apoptotic cells. Some macrophage receptors, such as CD36, might be able to directly interact with PS on the surface of apoptotic cells (18). In addition, several bridging molecules have been described as binding to PS on apoptotic cells, as follows: the milk fat globule protein MFG-E8 interacting with the vitronectin receptor αvβ3 (19, 20), the growth arrest–specific gene product GAS-6 representing a ligand for the tyrosine protein kinase receptor MER (6), β2-glycoprotein I binding to its receptor (21), and annexin I binding to its receptor (22). The interactions between phagocyte receptors and dead cells can be studied using mAb and other reagents that bind to individual components of the recognition and uptake machinery for apoptotic cells, and thereby interfere with the engulfment of dead cells (23).

When the clearance process is defective, apoptotic cells eventually lose their membrane integrity and proceed to secondary necrosis. This results in the uncontrolled leakage of noxious contents and potential autoantigens from the dying cells. To date, the mechanisms of phagocytic clearance have been best studied for apoptotic cells. In contrast, little is known about the receptors and ligands involved in the clearance of necrotic cells. We have previously shown that the binding of complement is mainly focused on primary necrotic cells and apoptotic cells progressing to secondary necrosis (14). Induction of necrosis by ATP-depleting agents leads to efficient uptake of the dying cells by human monocyte-derived macrophages (HMDMs) immediately after the integrity of the plasma membrane is lost. These “scheduled necrotic” cells are taken up even more readily than apoptotic ones (24). An early sign of apoptosis and a signal for recognition and uptake of apoptotic cells is the exposure of PS on the outer leaflet of the cytoplasmic membrane. Since PS is also exposed on necrotic cells, it was not surprising that PS-dependent mechanisms contribute to the uptake process of necrotic cells as well (24, 25). In addition, mannose-binding lectin (MBL) and pulmonary SP-A and SP-D bind to late apoptotic (i.e., secondary necrotic) and primary necrotic cells and facilitate their engulfment (13, 26).

Necrotic cells initiate proinflammatory responses and can mediate maturation of dendritic cells, leading to efficient antigen presentation (27). In contrast, apoptotic cells, which maintain their membrane integrity, exert strong antiinflammatory effects on monocytes and macrophages. In monocytes, the antiinflammatory signal is mainly mediated by engagement of CD36 and leads to inhibition of tumor necrosis factor (TNF), interleukin-1β (IL-1β), and IL-12 production, and an increase of IL-10 production (1). In macrophages, engagement by apoptotic cells of both CD36 and, presumably, other surface receptors, including one recognized by mAb 217G8E9, inhibits secretion of IL-1β, TNF, macrophage inflammatory protein 1α (MIP-1α), and IL-12, but induces prostaglandin E2 and transforming growth factor β (10, 28).

In the present study, we investigated whether phagocyte receptors known to be involved in the engulfment of apoptotic cells contribute to the uptake of primary necrotic cells. We used HMDMs as phagocytes and autologous primary necrotic peripheral blood lymphocytes (PBLs) generated by treatment with heat, methanol, or ethanol as prey. Our data show that the complement component C1q, TSP, the TSP receptor CD36, the vitronectin receptor (αvβ3), CD14, and a surface receptor recognized by mAb 217G8E9 are involved in the uptake of primary necrotic cells. In addition, we demonstrate that necrotic cells may have the capacity to inhibit proinflammatory cytokine secretion from macrophages stimulated with lipopolysaccharide (LPS).

MATERIALS AND METHODS

Cells.

Venous blood was drawn from normal healthy volunteers according to institutional guidelines. The blood was anticoagulated by the addition of 20 units/ml heparin. Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density-gradient centrifugation (Lymphoprep; Gibco Invitrogen, Karlsruhe, Germany). Residual platelets were removed by density-gradient centrifugation through a cushion of fetal calf serum (FCS). After 3 washes with phosphate buffered saline (PBS; Gibco Invitrogen), the cells were suspended at 8 × 106 cells/ml in Dulbecco's modified Eagle's medium (DMEM; Gibco Invitrogen), and 0.5 ml of the cell suspension was added per well of 48-well culture plates. After 2 hours of incubation at 37°C in 7.5% CO2, the cell layer was washed twice with PBS to remove nonadherent PBLs. The adherent monocytes were cultured in DMEM containing 10% FCS and 5 ng/ml recombinant human granulocyte–macrophage colony-stimulating factor (GM-CSF) plus 100 units/ml penicillin, 100 mg/ml streptomycin, and 200 mML-glutamine. Monocytes were cultured at 37°C in 7.5% CO2 for 8 days to generate HMDMs.

To determine whether the results are dependent on the way the macrophages were generated, alternative methods to differentiate HMDMs were used. Human monocytes were enriched by adherence as described above, but PBMCs were centrifuged through a cushion of autologous plasma instead of FCS, and macrophages were differentiated from monocytes in the presence of 20% autologous serum without the addition of any further cytokines/growth factors. Alternatively, human monocytes were purified from PBMCs using a monocyte isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany). The monocytes, isolated via the magnetic-activated cell sorting system, were suspended at 6 × 105 cells/ml in a medium containing 10% FCS and 5 ng/ml GM-CSF. Aliquots of 0.5 ml per well of the cell suspension were plated in 48-well culture plates and cultured as described above. As a third isolation method, we purified monocytes from EDTA-treated blood by Dextran sedimentation of erythrocytes and density-gradient centrifugation of the leukocyte-rich plasma using NycoPrep, according to the instructions of the manufacturer (Nycomed Pharma, Oslo, Norway). Similar results were obtained with each method.

Sera and serum components

To prepare autologous active human serum, whole blood was allowed to clot at room temperature for 30 minutes. After centrifugation at 900g for 20 minutes, serum was removed and stored at −80°C. Heat-inactivated autologous serum was generated by incubating active human serum at 56°C for 30 minutes. This treatment inactivates components of both the classical (C1 and C2) (29) and the alternative (factor B) (30) complement activation pathways. C1q-depleted serum (C1q-DS) was obtained from Innogenetics (Heiden-Westfalen, Germany). C1q was obtained from Sigma (Taufkirchen, Germany). All sera and serum components used were of human origin.

Carboxyfluorescein succinimidyl ester (CFSE) staining and induction of apoptosis and necrosis.

PBLs were obtained from the same donor as the monocytes, and stained with CFSE-diacetate (CFSE-DA; Molecular Probes, Leiden, The Netherlands). Approximately 107 PBLs in 10 ml PBS were incubated with 30 μl of CFSE-DA for 15 minutes at 37°C under 7.5% CO2, and then washed twice with 20 ml DMEM containing 10% heat-inactivated FCS. For the induction of apoptosis, PBLs were irradiated with ultraviolet B (120 mJ/cm2) and then incubated for 18 hours in DMEM supplemented with 10% heat-inactivated FCS. Less than 10% of the apoptotic cells were positive for trypan blue staining. For induction of necrosis in CFSE-DA–labeled PBLs, 3 different methods were used: incubation at 56°C for 30 minutes, treatment with 50% methanol on ice for 15 minutes, or treatment with 10% ethanol at 37°C for 1 hour. Cells were washed 3 times to remove residual methanol or ethanol. After induction of necrosis, cells were maintained for ∼30 minutes at 37°C. Necrosis was verified in each individual experiment before addition of the dead cells to the macrophages. More than 90% of the cells were positive for trypan blue, as well as double positive for propidium iodide (PI) and annexin V–fluorescein isothiocyanate (FITC) (Figure 1a).

Figure 1.

Analysis by flow cytometry of the uptake of necrotic peripheral blood lymphocytes (PBLs) by human monocyte-derived macrophages (HMDMs). Necrosis induction by heat treatment resulted in >90% of the cells being double positive for annexin V (AxV) and propidium iodide (PI) (a). Necrotic PBLs were labeled with carboxyfluorescein succinimidyl ester (CFSE) diacetate and HMDMs were labeled with DiI, and dot plots of the stained necrotic PBLs (b) and stained macrophages (c) are displayed. Macrophages (gated on large cells in the forward scatter) that had taken up necrotic material (d) or apoptotic material (e) became double positive. The ingestion of the necrotic material was confirmed by confocal laser microscopy (f). Uptake of primary necrotic cells was clearly observed in the presence of autologous serum and in the absence of an inhibitor (g). Addition of antithrombospondin antibodies (20 μg/ml) markedly decreased the percentage of macrophages containing necrotic cells (h). FITC = fluorescein isothiocyanate.

Assay for phagocytosis of dying cells by macrophages.

To monitor the uptake of the CFSE-DA–stained necrotic or apoptotic cells into macrophages, we used 2-color flow cytometry. For this purpose, we stained the phagocytes with the membrane dye DiI (Molecular Probes, obtained via MoBiTec, Gottingen, Germany) and washed twice with DMEM. The dead cells were labeled with CFSE-DA (see above). For inhibition experiments, macrophages were preincubated for 15 minutes with 20 μg/ml anti-TSP (purified from the hybridoma cell line αhTSP-1; American Type Culture Collection, Manassas, VA) (31), 10 μg/ml anti-CD36 (Immunotech, Hamburg, Germany), 40 μg/ml mAb 217G8E9 directed against a putative PS receptor (Alexis Biochemicals, Lausanne, Switzerland), 40 μg/ml anti-CD14 (both kindly provided by Dr. J. Donald Capra, Oklahoma Medical Research Foundation, Oklahoma City) (32, 33), or with corresponding concentrations of isotype-matched control antibodies (Sigma, Deisenhofen, Germany, or own production), 2 μg/ml synthetic cyclic peptide Arg-Gly-Asp (RGD), or control peptide (kindly provided by Dr. L. Sorokin, Lund University, Lund, Sweden). After preincubation, autologous necrotic or apoptotic PBLs were added at approximately a 5-fold excess over the number of HMDMs. Alternatively, the dying cells were preincubated with 5 μg/ml human annexin V to block PS before adding them to the macrophages.

Antibody preparations containing LPS (tested by Limulus amebocyte assay) or sodium azide were purified using polymyxin B columns (Perbio Science, Bonn, Germany) or ultrafiltration, respectively. To further exclude LPS contamination during monocyte isolation and culture, we used only certified endotoxin-free one-way plastic articles; commercial media and PBS were also certified to be endotoxin free. Autologous serum was prepared using endotoxin-free sterile containers.

For the phagocytosis assays, 8-day-old HMDMs were labeled with DiI and coincubated with CFSE-labeled apoptotic or necrotic cells for 60 minutes at 37°C under 7.5% CO2. After labeling and incubation, the monolayer of HMDMs was washed with PBS to remove uningested dead PBLs. HMDMs were detached using Accutase (Innovative Cell Technologies, San Diego, CA) for 15 minutes at 37°C, transferred to tubes, and fixed with PBS containing 1% paraformaldehyde.

To quantify the uptake of the CFSE-labeled apoptotic or necrotic cells into DiI-labeled macrophages, we used 2-color flow cytometry: double-positive (carboxyfluorescein detected in fluoresence channel 1 [FL1] and DiI detected in FL2) macrophages had taken up the dead “prey.” In order to combine the results from multiple experiments with macrophages from different donors displaying marked donor-to-donor variations in the basal uptake level, we normalized for the differences in individual basal uptake levels using a phagocytosis index. First we electronically gated on HMDMs, which were identified by size using forward scatter/side scatter analysis and by positive staining for DiI, as detected in FL2. Within the gated population of HMDMs, the fluorescence intensity in FL1 indicated the amount of phagocytosed CFSE-labeled dead cells. The increase in the mean fluorescence intensity of FL1 after incubation of HMDMs with CFSE-labeled necrotic PBLs in the presence of control sera (heat-inactivated autologous serum or C1q-DS) was regarded as 100%. The phagocytic index was defined as the increase in the mean fluorescence intensity (FL1) for HMDMs incubated with CFSE-labeled necrotic PBLs in the presence of active human serum or C1q-reconstituted serum, and the increase in the mean fluorescence intensity (FL1) for HMDMs incubated with CFSE-labeled necrotic PBLs in the presence of the corresponding control sera (heat-inactivated autologous serum or C1q-DS), multiplied by 100. If inhibitors were used, the phagocytic index was defined as the increase in the mean fluorescence intensity (FL1) for HMDMs after incubation with CFSE-labeled necrotic PBLs in the presence of an inhibitor divided by the increase in the mean fluorescence intensity (FL1) for HMDMs after incubation with CFSE-labeled necrotic PBLs in the absence of the inhibitor (control), multiplied by 100. Laser scanning microscopy (Leitz, Wetzlar, Germany) was used to confirm the intracellular location of CFSE-labeled apoptotic cells within HMDMs in order to validate our flow cytometry–based assay.

Determination of cytokine secretion pattern.

Monocytes were prepared using Dextran sedimentation and NycoPrep purification, as described above. The purity of monocytes was analyzed by flow cytometry and routinely resulted in>85% CD14+ cells. Monocytes were resuspended at a concentration of 0.8 × 106/ml in RPMI 1640/Glutamax medium (Gibco Invitrogen) containing 10% heat-inactivated FCS. The cells were transferred into 48-well culture plates (900 μl/well) just prior to the addition of apoptotic or necrotic autologous lymphocytes. Alternatively, 8-day-old HMDMs differentiated from adherent monocytes cultured in DMEM containing 20% autologous serum were used. Living, apoptotic, and necrotic cells (1.0–1.2 × 106/well) were added to the monocytes or macrophages, respectively, in 50 μl of medium 1 hour prior to activation with LPS (100 ng/ml). Supernatants were harvested 16 hours after LPS activation. Cytokine concentrations were determined by enzyme-linked immunosorbent assay, using appropriate pairs of mAb specific for human TNF and IL-10 (BD PharMingen, Heidelberg, Germany). The experiments were also performed in the same way using macrophages incubated with washed and unwashed primary necrotic cells.

Statistical analysis.

Statistical analyses were performed using Student's two-tailed t-test. P values less than or equal to 0.05 were considered significant; P values less than or equal to 0.01 were considered highly significant.

RESULTS

Quantitative monitoring of HMDM uptake of necrotic cells by flow cytometry.

Necrotic cell death of PBLs was induced by incubation at 56°C for 30 minutes. Before addition to macrophages, >90% of the cells displayed a necrotic phenotype and were double positive for PI and annexin V (Figure 1a), as well as for trypan blue. In additional experiments, necrosis was induced either by treatment with 50% methanol on ice for 15 minutes or by treatment with 10% ethanol at 37°C for 1 hour, in order to investigate the influence of different methods of necrosis induction on the phagocytosis of dead cells. Phagocytosis tests in the presence of the various inhibitors revealed that there was virtually no difference in the receptor systems used for the uptake of necrotic PBLs induced by heat, ethanol, or methanol treatment (data not shown).

To investigate whether the methods used for isolation of monocytes and their differentiation into macrophages influence their mechanisms for necrotic cell uptake, we performed phagocytosis assays with macrophages generated by various methods. Monocytes purified by adherence or negative selection using magnetic beads behaved very similarly in the phagocytosis assays. Also, similar results were obtained with macrophages cultured in FCS and GM-CSF (Figure 2) and those cultured in 20% of autologous serum (Figure 3 and data not shown).

Figure 2.

Uptake of necrotic peripheral blood lymphocytes (PBLs) (generated by heat treatment) by human monocyte-derived macrophages (HMDMs) (isolated by adherence and differentiated in the presence of fetal calf serum and granulocyte–macrophage colony-stimulating factor), expressed as a phagocytosis index. The phagocytosis by HMDMs of necrotic PBLs was dependent on the thrombospondin–CD36–αvβ3 system and on CD14. Note the decreased phagocytic capacity of macrophages in the presence of anti-CD36 antibody, antithrombospondin antibody (anti–hTSP-1), or cyclic RGD peptides (a), and the decreased phagocytic capacity of macrophages in the presence of anti-CD14 antibody (b). Assays were performed in medium containing 10% heat-inactivated fetal calf serum. Values are the mean ± SD from ≥7 independent experiments. ∗ = P < 0.05; ∗∗ = P < 0.01. CO = control; Ab = antibody.

Figure 3.

Reduction of phagocytosis in HMDMs of primary necrotic cells. The uptake of necrotic PBLs (generated by heat treatment) by HMDMs, isolated by adherence and differentiated in the presence of 20% autologous serum, is shown as the percentage of phagocytes containing necrotic material. Note the decreased phagocytosis of macrophages in the presence of anti–hTSP-1, anti-CD36 antibody, or annexin V (AxV). The combination of anti–hTSP-1, anti-CD36, and annexin V resulted in only a slightly further reduction of the phagocytosis of primary necrotic PBLs. Values are the mean ± SD and are representative of 5 experiments, each performed in quadruplicate. ∗ = P < 0.05; ∗∗ = P < 0.01. See Figure 2 for other definitions.

In order to quantify the uptake of primary necrotic cells by HMDMs, we established a flow cytometric phagocytosis assay. As shown in Figure 1, the CFSE-labeled necrotic cells detected in FL1 (Figure 1b) could clearly be separated from macrophages that were stained with the lipophilic dye DiI detected in FL2 (Figure 1c).

Macrophages that had taken up necrotic cell material were positive for both DiI and CFSE (Figure 1d). Unlike the findings in phagocytosis assays with apoptotic cells (Figure 1e), we observed that with necrotic cells there were usually not such clearly distinct populations of macrophages that had taken up the cells and macrophages that had not taken up necrotic cells. This may be due to the fact that necrotic cells and their fragments display a higher variation in size and fluorescence intensity than apoptotic cells, since necrotic cells disintegrate during culture and their CFSE-staining intensity decreases. To demonstrate that the DiI/CFSE double-positive macrophages had ingested the dead cells, rather than just attached to their surface, we confirmed intracellular localization by confocal laser microscopy (Figure 1f). Detachment of adherent macrophages using Accutase virtually abolished necrotic cells attached to the surface of macrophages (results not shown). Therefore, for flow cytometric analysis, HMDMs were always detached using Accutase.

Inhibition of the uptake of primary necrotic cells could be detected using this flow cytometry–based phagocytosis assay. Figure 1g shows the uptake of primary necrotic cells in the absence of an inhibitor. In this representative experiment, the percentage of macrophages that had ingested necrotic cell material was decreased from 10.5% to 5.1% in the presence of anti–TSP-1 (Figure 1h). There was a relatively high variation in the basal uptake of necrotic and apoptotic cells by macrophages from different donors. To combine the results of different experiments, we normalized the various phagocytosis assays by applying a phagocytic index (see Materials and Methods).

Involvement of heat-labile serum components in the uptake of primary necrotic cells.

To evaluate the influence of heat-labile serum components on phagocytosis, we performed the phagocytosis assays with autologous serum (either heat-inactivated or active serum). As shown in Figure 4a, in the presence of active human serum the phagocytosis of apoptotic cells by HMDMs was significantly increased. Similarly, the uptake of primary necrotic cells, whether generated by heat, methanol, or ethanol treatment, was also markedly higher in the presence of active human serum compared with heat-inactivated autologous serum (Figure 4b). In general, the difference in the increase with active human serum versus heat-inactivated autologous serum was more pronounced for phagocytosis of necrotic cells than for phagocytosis of early apoptotic cells, which are still PI negative (Figure 4 and data not shown), indicating that heat-labile serum components such as certain complement proteins are of special importance for the removal of necrotic cells.

Figure 4.

Phagocytosis of apoptotic and primary necrotic peripheral blood lymphocytes (PBLs) by human monocyte-derived macrophages (HMDMs) in the presence of autologous active serum (AS) versus heat-inactivated serum (IS), and the role of C1q in the uptake of primary necrotic cells. The uptake of dying PBLs by HMDMs (isolated by adherence) is expressed as a phagocytosis index. Note the increased phagocytic capacity for the uptake of both apoptotic (a) and necrotic (b) cells by HMDMs in the presence of active human serum. The addition of C1q to heat-inactivated serum (c) led to a slight increase and the reconstitution of C1q-depleted serum (C1q-DS) with C1q (d) led to a strong increase of the phagocytic capacity for the uptake of necrotic cells by HMDMs. Values are the mean ± SD from 7 independent experiments in a and b and 4 independent experiments in c and d. ∗ = P < 0.05; ∗∗ = P < 0.01.

Augmentation of the phagocytosis of primary necrotic cells by C1q.

We recently demonstrated that the heat-labile complement component C1q augments the DNase I–mediated degradation of necrotic cell–derived chromatin and is necessary for effective uptake of necrotic cell–derived chromatin by activated monocytes (34). Regarding the uptake of necrotic cells by professional phagocytes such as macrophages, we found that C1q markedly contributed to the clearance of whole necrotic cells. Adding physiologic concentrations (25 μg/ml) of heat-labile C1q to serum-free medium (data not shown) or heat-inactivated autologous serum (Figure 4c) increased the phagocytosis of necrotic cells by HMDMs only slightly. However, reconstitution of C1q-depleted serum with C1q significantly elevated the phagocytosis of necrotic cells (Figure 4d), indicating that other heat-labile serum components, such as other complement proteins, may also be involved in the C1q-mediated uptake process.

Importance of PS for the uptake of primary necrotic cells by HMDMs.

The exposure of PS on the outer leaflet of apoptotic and necrotic cells is an important primary signal recognized by phagocytes (24, 25). We used HDMDs to examine whether the uptake of autologous primary necrotic PBLs is dependent on PS. For this purpose, we performed blocking experiments using the protein annexin V, which binds to PS with high affinity in a calcium-dependent manner. In the presence of annexin V, we observed a significant reduction in the uptake of heat-treated necrotic cells by HMDMs (Figure 5). This inhibition was also observed with annexin V–preincubated primary necrotic cells generated by methanol or ethanol treatment (data not shown). These data indicate that the PS-dependent uptake of primary necrotic cells is independent of the necrosis-inducing stimuli.

Figure 5.

Phagocytosis by human monocyte-derived macrophages (HMDMs) of necrotic peripheral blood lymphocytes (PBLs) is dependent on phosphatidylserine (PS) and a surface receptor recognized by monoclonal antibody (mAb) 217G8E9. The uptake of necrotic PBLs (generated by heat treatment) by HMDMs (isolated by adherence) is expressed as a phagocytosis index. Note the decreased phagocytic capacity of macrophages in the presence of PS-binding annexin V (AxV) and mAb 217G8E9, respectively. Assays were performed in medium containing 10% heat-inactivated fetal calf serum. Values are the mean ± SD from ≥4 independent experiments. ∗∗ = P < 0.01. Co = control; Ab = antibody.

To further investigate the PS-dependent uptake of primary necrotic cells, we used the monoclonal antibody mAb 217G8E9, which appears to block the PS-dependent uptake of apoptotic cells and is believed to be directed against a not-yet-identified PS receptor (35). In the presence of mAb 217G8E9, the phagocytosis of necrotic cells was significantly reduced in comparison with isotype-matched control antibodies (Figure 5). Therefore, the surface receptor recognized by mAb 217G8E9 contributes to the uptake of not only apoptotic cells but also necrotic cells.

The role of the thrombospondin–CD36–αvβ3 system for the uptake of primary necrotic cells by HMDMs.

Thrombospondin 1, which binds to apoptotic cells, forms a “molecular bridge” to the TSP receptor CD36 and its coreceptor αvβ3 on macrophages, thereby mediating the uptake of apoptotic cells (2, 11). However, it remained unclear whether the thrombospondin–CD36–αvβ3 system is involved in the uptake of necrotic cells as well. When anti–TSP-1 or anti-CD36 antibodies were included in phagocytosis assays, we found a significantly reduced uptake of heat-treated primary necrotic cells (Figure 2a). Blockade of the vitronectin receptor αvβ3 using cyclic RGD peptides, but not scrambled cyclic peptides, also resulted in an effective inhibition of the uptake of primary necrotic cells generated by heat treatment (Figure 2a). Virtually identical results were observed if necrotic cells were produced using the methanol or ethanol treatment methods (data not shown). Anti-CD36, anti–TSP-1, and cyclic RGD peptides were similarly efficient in blocking the phagocytosis of apoptotic and necrotic cells (data not shown).

Role of CD14 in the uptake of primary necrotic cells by HMDMs.

It has been previously reported that the plasma-membrane glycoprotein CD14 on the surface of human macrophages is involved in the recognition and clearance of apoptotic cells (33). In our phagocytosis assays, preincubation of HMDMs with the anti-CD14 antibody 61D3, which has been shown to inhibit the phagocytosis of apoptotic cells (32, 33), significantly reduced the engulfment of autologous heat-treated necrotic lymphocytes (Figure 2b).

Influence of combined blockade of the thrombospondin–CD36–αvβ3 system and PS on the uptake of primary necrotic cells by HMDMs.

We applied different isolation and differentiation methods to be sure that the results obtained were independent of the monocyte isolation and macrophage differentiation. Figure 3 shows the results of 1 representative experiment, in which HMDMs were differentiated in the presence of 20% autologous serum without addition of cytokines. The combination of anti-CD36, anti–TSP-1, and annexin V led to only a slight further reduction of the phagocytosis of primary necrotic cells compared with the individual inhibitors. Although this slightly stronger inhibition of phagocytosis by combining various blockers was reproducible in 2 independent experiments, it did not reach statistical significance. This finding is consistent with earlier data indicating that CD36, αvβ3, and TSP belong to the same uptake pathway. There was also only a subtle, nonsignificant further decrease in the uptake of necrotic cells if annexin V was used to block PS, which appears to be involved in additional recognition/uptake pathways. In summary, in the absence of heat-labile serum components, blockade of the thrombospondin–CD36–αvβ3 system and PS-dependent mechanisms reduced the percentage of phagocytes containing necrotic cell material by ∼50% (Figure 3).

Ligation of antiinflammatory receptors by necrotic cells influences phagocyte function.

Ligation of CD36 and a surface receptor recognized by mAb 217G8E9 has been shown to induce an antiinflammatory state in monocytes and macrophages, respectively. Our blocking experiments using anti-CD36 and mAb 217G8E9 indicated that necrotic cells also engage CD36 and the putative surface receptor defined by mAb 217G8E9. Therefore, we tested whether necrotic PBLs may be able to deliver an antiinflammatory signal to phagocytes, although usually the proinflammatory effects of necrotic cells may dominate. As shown in Figure 6, necrotic PBLs reduced the LPS-induced secretion of TNF and increased the secretion of IL-10 in primary human monocytes (Figure 6a) and in HMDMs (Figure 6b), respectively. However, this antiinflammatory effect was usually weaker than the antiinflammatory effect of apoptotic cells (Figure 6a and data not shown).

Figure 6.

Influence of apoptotic (apo) and primary necrotic (necro) cells on cytokine secretion pattern of lipopolysaccharide (LPS)–stimulated monocytes and macrophages. Interleukin-10 (IL-10) and tumor necrosis factor α (TNFα) secretion by purified human monocytes that were stimulated with LPS in the presence of living, apoptotic, or washed primary necrotic cells was quantified by enzyme-linked immunosorbent assay (a). The same experiments were performed with human monocyte-derived macrophages (HMDMs) isolated by adherence, using unwashed heat-induced necrotic peripheral blood lymphocytes (PBLs) (b). Values are the mean ± SD from 4 assays. ∗∗ = P < 0.01 versus untreated control medium.

We also performed experiments using macrophages incubated with unwashed (Figure 6b) and washed primary necrotic cells. In this system with heat-induced necrosis of primary lymphocytes, there was no significant difference between unwashed and washed necrotic cells. In earlier experiments using necrotic granulocytes, we observed a slight induction of proinflammatory cytokines by unwashed, but not washed, necrotic granulocytes in the absence of LPS stimulation. In these experiments using lymphocytes as necrotic cells, there was no detectable induction of cytokines observed in the absence of LPS stimulation (Figure 6b and data not shown). Therefore, the proinflammatory properties of necrotic cells may also depend on the origin of necrotic cells, which may contain differing amounts of mainly soluble proinflammatory mediators.

DISCUSSION

If apoptotic cells, which usually maintain their membrane integrity for a relatively long time, are not properly cleared, they reach a state of secondary necrosis, and autoimmune reactions against components of the lysed cells may arise. During the process of apoptosis, proteins are often modified or cleaved at unusual sites, resulting in neoantigens, which may contribute to break self-tolerance. It is notable that most autoantigens in systemic lupus erythematosus (SLE) can be cleaved by the action of granzyme B, which is secreted by cytotoxic T lymphocytes in order to kill virus-infected or other target cells (36, 37). Therefore, impaired engulfment of dead cells releasing modified or unmodified potential autoantigens is believed to play a key role in the etiopathogenesis of autoimmune diseases such as SLE (38–40).

SLE is a multifactorial disease, and its pathogenesis and precise etiology remain elusive. Neither apoptotic cells, necrotic cells, nor dying-cell–derived material (e.g., chromatin) can be easily found in issues under physiologic conditions, due to fast and efficient removal by a highly effective scavenger system. Autoantigens are found in apoptotic and necrotic cells and are recognized by autoimmune sera from SLE patients. Dying cells are cleared via a redundant system of receptors on phagocytic cells and bridging molecules, which detect molecules specific for dying cells. Changes on the surfaces of apoptotic and necrotic cells are extremely important for their recognition and disposal. Some SLE patients appear to have an impaired ability to clear apoptotic and necrotic cells from tissues; this could predispose to disruption of mechanisms of peripheral tolerance against self antigens (39, 41, 42).

In vivo cells undergoing apoptotic cell death are cleared rapidly by phagocytes without induction of inflammation. In late stages of apoptosis, dead cells can be cleared via backup mechanisms such as complement-mediated uptake (12, 14, 41, 43). Recently, it has been shown that the high mobility group box chromosomal protein 1 (HMGB-1), which is “frozen” on the chromatin of apoptotic cells and remains immobilized even under conditions of secondary necrosis, is released by primary necrotic cells and contributes to the inflammatory response (44). Therefore, primary and secondary necrotic cells may induce different inflammatory and antiinflammatory signals and, thereby, govern the immunologic responses.

Hirt and Leist showed that “scheduled” necrotic cells generated by ATP-depleting conditions, in contrast to “unscheduled” primary necrotic cells, reduced the release of proinflammatory cytokines from macrophages (24). Conversely, Brouckaert and colleagues reported that uptake of apoptotic or necrotic L929 cells by macrophages did not modulate the expression pattern of proinflammatory cytokines by macrophages at the messenger RNA and protein level (25). Our results clearly indicate that necrotic cells, like apoptotic cells, also can engage CD36 and a surface receptor recognized by mAb 217G8E9, and thereby mediate antiinflammatory signals (Figures 2a, 3, 5, and 6). These antiinflammatory signals may usually be overcome by the release of proinflammatory mediators such as heat-shock proteins and HMGB-1. Nevertheless, there might be conditions such as “scheduled necrosis” in which the antiinflammatory signals may dominate. Once necrotic cells have released their “danger signals” they might be phagocytosed without causing further inflammatory responses. Washing removes the released proinflammatory mediators, whereas surface structures engaging the thrombospondin–CD36–αvβ3 system or the putative PS receptor are unaffected. These findings may have implications for development of cancer vaccines based on primary necrotic cells.

In experiments using necrotic granulocytes, we observed a slight induction of proinflammatory cytokines by unwashed, but not washed, necrotic granulocytes in the absence of LPS stimulation (data not shown). In the experiments shown in Figure 6b using lymphocytes as necrotic cells, there was no detectable induction of cytokines observed in the absence of LPS stimulation. We conclude that the proinflammatory properties of necrotic cells depend also on the origin of necrotic cells, which may contain different amounts of (mainly soluble) proinflammatory mediators.

Important receptors on monocyte/macrophages mediating the antiinflammatory clearance of apoptotic cells are the thrombospondin–CD36–αvβ3 receptor system (1, 11) and, presumably, a receptor engaged by mAb 217G8E9 (28, 45, 46). Two groups of investigators have reported phenotypes of PS receptor–deficient mice that appeared consistent with the notion of a role of the PS receptor in the removal of apoptotic cells (47, 48). However, the role of the mainly nuclear protein termed PS receptor in mediating the phagocytosis of dead cells has been called into question by results of a recent study in which the phenotype of another PS receptor–deficient mouse line was carefully investigated. Importantly, mAb 217G8E9 still reacted with the surface of macrophages derived from PS receptor–deficient mice (35). Nevertheless, a not-yet-identified receptor on macrophages for the PS-dependent clearance of dead cells, which can be blocked by mAb 217G8E9, may play an important role in the uptake process of apoptotic as well as necrotic cells.

In the present study, we showed that several recognition systems known to be involved in the clearance of apoptotic cells, such as complement-mediated mechanisms, a surface receptor recognized by mAb 217G8E9, the thrombospondin–CD36–αvβ3 receptor system, and CD14, also contribute to the clearance of primary necrotic cells. Our results are consistent with findings in the nematode Caenorhabditis elegans, in which a common set of engulfment genes mediates the uptake of both apoptotic and necrotic-like cells (49). Since the phagocytosis process is highly conserved from nematodes to humans, injured necrotic cells in higher organisms may also be at least partially eliminated through a very well-regulated process of dead cell removal largely common for apoptotic and necrotic cells (49).

In contrast to TNF-induced necrotic cells (25), necrotic cells generated by ATP depletion (“scheduled necrosis”) were taken up more efficiently than apoptotic ones. Alternatively, these differences might be due to different phagocytes used in these studies. The uptake of these “scheduled necrotic” cells was independent of the CD36 and the CD14 receptor systems (24). Conversely, we show that the thrombospondin–CD36–αvβ3 system, CD14, and the receptor recognized by mAb 217G8E9 all contribute to the process of clearance of autologous primary necrotic lymphocytes by HMDMs.

In the first report describing a role of CD14 in the uptake of apoptotic cells, it was mentioned that the anti-CD14 antibody 61D3 did not significantly inhibit the uptake of necrotic cells (33). The partial divergence of these data from our present results may be due to different experimental systems, and may indicate that clearance efficiency and pathways might depend on the origin of necrotic cells, necrosis-inducing conditions, subsets of phagocytes, or their maturation state. For instance, in the presence of active human serum, necrotic, but not apoptotic, lymphocytes are efficiently removed via C1q-dependent mechanisms, thereby presumably partially compensating for the blockade of CD14 and CD36. Therefore, we performed the blocking experiments with anti-CD14 mAb 61D3 and inhibitors of the thrombospondin–CD36–αvβ3 system in heat-inactivated FCS, enabling us to clearly detect the contribution of CD14 and CD36 to the uptake of necrotic cells. Heasman and colleagues showed that the cytokine microenvironment at sites of inflammation will critically determine the functional capacity of monocytes following treatment with glucocorticoids (50). However, in our experiments, monocytes isolated by adherence or negatively selected using magnetic beads and differentiation with or without GM-CSF showed no major influence on uptake systems used for phagocytosis of necrotic cells.

We have described previously that early complement components of the classical pathway are involved in the clearance of very late apoptotic cells (14). We also demonstrated that the complement component C1q is also of major importance for the uptake of primary necrotic cells by HMDMs. Adding physiologic concentrations of C1q (25 μg/ml) to serum-free medium (data not shown) or medium containing 10% heat-inactivated FCS (Figure 4c) led to only a slight increase of the phagocytosis index, indicating that C1q alone is not sufficient for mediating the phagocytosis of dead cells. However, the uptake of primary necrotic cells was markedly increased by adding C1q to C1q-depleted serum, which still contains other potentially important serum components (Figure 4d). Quartier and colleagues also showed that C1q alone at a concentration of 25 μg/ml displayed only very little influence on phagocytosis of apoptotic cells, whereas C1q at higher concentrations (100 μg/ml), or the addition of soluble IgM, resulted in a significant increase of the phagocytosis of apoptotic thymocytes by murine bone marrow–derived macrophages (51).

Taken together, our data indicate that several receptor systems for the removal of apoptotic cells are also used for the clearance of necrotic cells. However, secondary and primary necrotic cells may use additional removal systems, such as complement, which may be of particular importance for the clearance of late apoptotic and necrotic cells. Thus far, we have not identified a receptor that is exclusively specific for apoptotic cells. It is likely that additional receptors contribute to the uptake of necrotic cells; some may specifically recognize necrotic cells, and others may bind to both apoptotic and necrotic cells.

Many different kinds of recognition and uptake systems are involved in the removal of primary and secondary necrotic cells. Among patients with SLE, ∼40–50% show evidence of impaired clearance of apoptotic as well as necrotic cells, which may be the main reason for the accumulation of extracellular nuclear material (40). Various alterations to the receptor systems for dead cells may, therefore, predispose individuals to autoimmune diseases such as SLE. We are currently continuing to investigate the reasons for the defective clearance in SLE, and also testing the receptor systems for dead cells. How the clearance efficiency of distinct dying cell populations modifies the immune response, and the release of autoantigens, and potentially contributes to breaking self tolerance, will be of major interest for the understanding of the etiopathogenesis of SLE.

Acknowledgements

We thank Dr. Dirk Mielenz for his valuable assistance with confocal microscopy. The anti-CD14 antibody was kindly provided by Dr. J. Donald Capra. The synthetic cyclic peptide Arg-Gly-Asp (RGD) and the control peptide were kindly provided by Dr. L. Sorokin.

Ancillary